Parallel corpora and semantic change 123
with similar literal or canonical core senses in L1 and L2 (as we have seen in
Example (11) where quand même is translated though).- The L1 term may be polysemous – but so, too, may the L2 term. Which of the
many possible senses of the range of L2 terms should we read as the interpre-
tation of the L1 term?
With particular reference to quand même and the possibility of exploring the
semantic evolution of pragmatic markers more generally through translation
equivalence, there is the specific problem of the lack of translated spontaneous col-
loquial language. Pragmatic markers are known to be characteristic of the spoken
language and to appear far more rarely in written text. One of the best resources
might appear to be film subtitles but film scores are generally scripted and writers
rarely include pragmatic markers.
Parallel corpora can, however, complement evidence gathered from other
sources – and translations have the decided advantage of combining insights from
a number of different interpreters of meaning, engaged in authentic communica-
tive behaviour.
Corpora consulted
Spoken
ESLO (1966–1970) Enquête Sociolinguistique d’Orléans – 109 hours of spoken French (902,755
words transcribed)
http://bacharts.kuleuven.ac.be/elicop
BC (1988) Beeching Corpus – 17.5 hours of spoken French, 155,000 words transcribed, 95
speakers.
http://www.uwe.ac.uk/hlss/llas/iclru/corpus.pdf
CRFP (2002) Corpus de Référence du Français Parlé – 40 towns in France, 400,000 words.
Available via a concordancer at
http://sites.univ-provence.fr/delic/corpus/index.html
CFPP (2000–) Corpus du Français Parlé Parisien – 28 speakers, 36.6 hours of spoken interviews,
transcribed and available.
http://cfpp2000.univ-paris3.fr/
Written
FRANTEXT – literary corpus: 210 million words in 3,737 texts from the 16th. to the 21st. cen-
tury. Subscription only. http://www.frantext.fr/